SENTIMENT ANALYSIS ON TWITTER HEALTH NEWS

Abstract

Microblogging has become a generally accepted way of expressing opinions and sentiments about products, services, media, institutions to mention but few. A lot of research has focused on analyzing Twitter health news for topic modelling using various clustering approaches, but few have reported it for sentiment analysis. The fact that such data contains potential information for revealing the opinion of people about health services and behaviours make it an interesting study. In this paper, general sentiments about Twitter health news was investigated. Natural language processing and text mining tool, AYLIEN API was used to
extract sentiments subjectivities and polarities from a previously uncategorized dataset. The result shows that most of the tweets in Twitter health news are objective, that is, expressing facts with an average of 64% objective while 34% are personal views or opinions (subjective) and subjectivity confidence of 0.9. Sentiment polarity reveals 9% positive, 19% negative and 72% neutral with polarity confidence of 0.6.